VOLUME 23
NUMBER 8
AUGUST 1990 Registered in US.Patent and Trademark Office; Copyright 1990 by the American Chemical Society
Electronic Structures and Reactivities of Metal Cluster Complexes WILLIAMC. TROGLER Department of Chemistry, 0 - 0 0 6 , University of California at San Diego, La Jolla, California 92093-0506 Received December 29, 1989 (Revised Manuscript Received March 28, 1990)
The effect of metal-metal interactions on the physical properties and reactivity of transition-metal cluster complexes has been an area of interest in our laboratory for over a decade. At the time we began these studies, the nature of metal-metal interactions in dinuclear complexes was well understood; most advanced undergraduate inorganic texts contain a discussion of metal-metal u, T , and 6 bonds. The degree of complexity increases rapidly as one considers clusters of three, four, and six metals so that a qualitative discussion of the bonding and its impact on physical properties becomes difficult. One question we hoped to address was whether an energy-band approach to small molecular clusters offered any advantage in understanding their electronic structures. Theoretical studies of pure metallic clusters suggest that as few as 40-80 metal atoms result in some properties similar to the bulk metal.' Another point of concern was the effect of delocalized metal-metal bonding on the thermal and photochemical reactivity of polynuclear complexes. Photochemical metal-metal bond homolysis in dinuclear complexes was well established and shown to result from excitation of a u u* electronic transition;2however, the situation for larger clusters was more ~ o m p l e x . ~ In particular, the factors that control metal-metal bond homolysis vs metal-ligand bond cleavage in both thermal and photochemical reactions were topics that interested us. Since polynuclear compounds often serve as delocalized pools of electrons for redox chemistry (e.g., ferredoxins),
-
Willlam C. Trogler, born on December 23, 1952, in Baltimore, MD, is a Professor of Chemistry at the University of California at San Diego. He received B.A. and M.A. degrees from the Johns Hopkins University and studied with Harry Qray at Caltech for his Ph.D. He was a member of the chemistry faculty at Northwestern University before Joiningthe department at San Diego. His research interests span physicaCmechanistic and catalytic aspects of coordination and organometallic complexes.
the consequences of delocalized bonding on cluster redox properties and reactivity provided another question to be examined. Theoretical Methods. Electron counting rules have proved to be extremely valuable for predicting cluster geometries and ~tabilities.~For our goal of understanding spectroscopicproperties, charge distributions, and reactivities, we sought a first principles theory capable of providing such information. To us the best compromise between theoretical rigor and the ability to perform computations in a reasonable time was the X a pr~cedure.~In this approximate method one replaces the exchange potential operator, Kj., in the Hartree-Fock (HF) equations with the expression of eq 1. This equation describes the exchange energy for a
free-electron gas, which depends only on the electron density distribution, p , and avoids the calculation of time-consuming exchange integrals. The parameter a( -0.7) can be determined by fitting atomic calculations to the HF results. It has been argued5 that this approximation incorporates correlation between electrons. This may have some validity in view of the success the X a method has shown in treating weak metal-metal bonds, such as in Cr2(02CH)4,6and in (1) Melius, C. F.; Upton, T. H.; Goddard, W. A., 111. Solid State Commun. 1978,238, 501. (2) Geoffroy, G. L.;Wrighton, M. S. Organometallic Photochemistry; Academic Press: New York, 1979. (3) Manning, M. C.;Trogler, W. C. Coord. Chem. Rev. 1981,38,89. (4) Wade, K.Transition Metal Clusters; Johnson, B. F. G., Ed.;Wiley: New York, 1981; pp 193-224. (5) Slater, J. C. The Self-Consistent Field for Molecules and Solids; McGraw-Hill: New York, 1974.
0001-4842/90/0123-0239$02.50/0 0 1990 American Chemical Society
Trogler
240 Acc. Chem. Res., VoZ. 23, No. 8, 1990
diatomic transition-metal specie^,^ when compared to more computer intensive HF-CI procedures. In the X a approximation several different approaches have been applied to the calculation of the Coulomb potential. In the muffin tin (MT) X a scheme5 one assumes a spherically symmetrical potential within touching spheres around each atom. Outside these spheres the potential is constant. This procedure, originally designed for solids: is a poor approximation for molecules with a strong bond (e.g., CO). The scattered wave (SW) method8 allows the potential spheres to overlap and thereby allows flexibility in the description of the bonding regions. This produces better results for molecular specie^.^ We adopted an alternative treatment called the discrete variational method (DV)l0and were fortunate to engage in a fruitful collaboration with an inventor of the techniquelo*"in applications to organometallic systems. In this procedure one computes the various integrals on a grid of points (usually 300 points/atom) distributed throughout the molecule. The degree of numerical accuracy can be increased by increasing the number of such points (at an expense in computation time). It is still convenient to expand the charge density in spherical harmonics about each atom, and there are several ways to do this. By the use of a spherical wave expansion from an orbital charge population (SCC) analysis12 one obtains results that we have found to be similar to those of the SW method in a number of trial calculations on mononuclear and dinuclear complexes.12 A more rigorous procedure is to use an analytical least-squares fit of the molecular charge density with expansion functions localized on atoms, as well as between bonds.13 This permits one to approach an exact solution of the X a Hamiltonian. An important distinction for X a orbital energies is that they represent the partial derivative of the total energy with respect to the occupation number for that orbital. Therefore, one cannot simply equate an X a orbital energy diagram with that for a HF calculation. Although this is a disadvantage, because X a orbital energies cannot be equated with ionization potentials, the problem is solved by the transition-state m e t h ~ d . ~ Here one performs the calculation of a hypothetical transition state for a spectroscopic process. For example, to calculate the ionization potential from orbital di, one calculates the energy, ei, for this orbital with one-half of an electron removed (Le., for the +0.5 ion). For an electronic transition one removes one-half of an electron from an occupied orbital and promotes it to the appropriate unoccupied orbital. These calculations converge rapidly once the ground-state calculation has been completed and have the advantage of reoptimizing the orbitals for each specific transition. Our experience in transition-metal systems has been that the calculated (6)
Cotton, F. A.; Stanley, G. G. Inorg. Chem. 1977,16, 2668.
(7) Delley, B.; Freeman, A. J.; Ellis, D. E. Phys. Reu. Lett. 1983, 50, 488.
(8)Johnson, K. H. Annu. Reu. Phys. Chem. 1975,26, 39. (9) Norman,J. G., Jr. J. Chem. Phys. 1974, 61, 4630. (IO) Ellis, D. E.; Painter, G. H. Phys. Reu. E 1970,2, 2887. (11) Trogler, W. C.; Desjardins, S. R.; Solomon, E. 1. Inorg. Chem. 1979,18,2131. Trogler, W. C.; Johnson, C. E.; Ellis, D. E. Inorg. Chem. 1981,20,980. Gross, M. E.: Trogler, W. C.; Ibera, J. A. Organometallics 1982,1,732. Manning, M. C.; Holland, G. F.; Ellis, D. E.; Trogler, W. C. J. Phvs. Chem. 1983,87.3083. Lee, S . W.; Trmler, - W. C. Inorg. Chem. 1990,29, 1659. (12) Rosen, A.; Elliie, D. E. J. Chem. Phys. 1976,65,3629. (13) Delley, B.; Ellis, D. E. J. Chem. Phys. 1982, 76, 1949.
14
13
I.?
11
io
9
8
7
6
eV
Figure 1. Valence photoelectron spectrum of ResC19 vapor at
800 K. Vertical lines represent predicted ionizations from SCFXu-DV calculations. Adapted from ref 14.
energies usually are within 1.0 eV of experimental values. This substantial error is about what one can realistically expect from calculations of many-electron clusters. Trinuclear Metal Cluster Complexes. The first systems we examined, Re3C&14and M3(CO)12 (M = Ru and OS),'^ were chosen because of their volatility and high symmetry (&). This permitted the measurement of gas-phase photoelectron spectra (PES) and simplified the calculations. At the time, it was fortunate that Dr. Joseph Berkowitz of Argonne National Laboratories allowed me to vaporize these compounds into his PES system. It was a challenge to obtain a spectrum before the insulating film of deposited cluster destroyed the resolution of the spectrometer (resulting in a 4-h disassembly and cleaning of the instrument); however, satisfactory spectra were eventually obtained. In Figure 1 is displayed the valence photoelectron spectrum of Re3(&, measured at about 800 K, with vertical lines for the predicted orbital ionization potentials from an X a calculation. The close correspondence between the peaks in the experimental spectrum and the calculated ionization potentials, as well as the agreement between the absolute energies for the net splitting of the valence electron ionizations (8.9-13.86 eV experiment vs 8.7-14.2 eV calculated), gave us confidence in the merit of the computational approach. A striking feature of the system was the predicted high stability of a cluster three-center a-bonding orbital, formed by 3d orbitals directed toward the center of the cluster, denoted a, in Figure 2. A subsequent PES and SCF-Xa-SW studylB of the same system and the bromide analogue arrived at substantially the same conclusions with the inclusion of relativistic corrections.l7 (14) Trogler, W. C.; Ellis, D. E.; Berkowitz,J. J. Am. Chem. SOC.1979, 101,5896. (15) Delley,B.; Manning, M. C.;Ellis, D. E.; Berkowitz,J.; Trogler, W. C. Inorg. Chem. 1982,21,2247.
(16) Buraten, B. E.; Cotton, F. A.; Green, J. C.; Seddon, E. A.; Stanley, G . G . J . Am. Chem. SOC.1980,102,955.
Metal Cluster Complexes
Acc. Chem. Res., Vol. 23, No. 8,1990 241
+4-
-
-------8 0;
+2
0;p
: *io
-
2
HOMO
0-
R A
-
0,P
-- -........._._ _ - .. -14 e' - 9 e"
____... ----___.__-.---3 a;__.--
e'a,
-2-
-4-
Figure 2. Symmetry combinations of metal d orbitals in a trinuclear metal cluster.
__-- -______._----__-_.__ -:::.-.---
EO" 81" 9 8f
.--. ::...--
I_
_.--*
osl(col,z
RuJCOI,~
In subsequent spectroscopic and theoretical studies uf M3(C0)12clusters, we analyzedls metal d-orbital interactions according to the diagram in Figure 2. As for Re3C19,the most stable metal-localized orbital (Figure 3) was the three-center-cluster a-bonding orbital, denoted ac in Figures 2 and 3. Unlike the case of the Figure 3. Orbital energy diagrams for Ru&C0),2 and OS&CO)~ rhenium cluster system, the a,* orbital was also occufrom SCF-Xor-DVcalculations. Reprinted with permission from ref 15. Copyright 1982 American Chemical Society. pied in the carbonyl systems, along with the degenerate cluster x orbitals, denoted xc and uc*. These were split about the two ulorbitals of the cluster that engage in x-back-bonding to the carbonyl groups. Filled metalmetal antibonding orbitals in electron-rich carbonyl clusters can still experience a net stabilization by uback-bonding to the carbonyl groups, as shown for the 14e'( ac*) orbital in Figure 4. Net metal-metal bonding in the cluster derives from the occupied 1%' (a,) orbital that resembles a cyclopropane-like edge-bonding orbital. This four-electron a-bonding model corresponds well with Pauling's valence-bond description'* of two edge bonds resonating among three triangular edges in these clusters. The quantitative agreement obtainedls between the calculated d-orbital splitting and that observed in PES spectra of R U ~ ( C Oand ) ~ ~O S ~ ( C O ) ~ ~ suggests that the model is substantially correct. It was known that both metal-metal bond cleavage and CO dissociation could occur in the M3(CO)12 clusters on irradiation with ultraviolet light.l9 Therefore, we focused attention on the lowest electronic transitions in these compounds: the dipole-forbidden loa: 6a4 one-electron excitation and the dipole-allowed 15e' Figure 4. Contour plot of the 14e'(ac*) orbital for R U & O ) ~ ~ 684 process. The loa: orbital possesses a complex
--
(17) Christiansan,P. A.; M e r , W. C.; Pitzer, K. S. Annu. Rev. Phys. Chem. 1986,36,407. (18) Pauling, L. R o c . Natl. Acad. Sci. U.S.A. 1976, 73,4290. (19) Desrosiers, M. F.; Wink, D. A.; Trautman, R.; Friedman, A. E.; Ford, P. C. J. Am. Chem. Soc. 1986,108,1917. Ford, P. C. J. Orgonomet. Chem. 1990,33,339. Bentsen, J. G.; Wrighton, M. S . J. Am. Chem. SOC. 1987,109,4518. Bentam, J. G.; Wrighton, M. 9. J.Am. Chem. SOC.1987, 109, 4530 and references therein.
Reprinted with permission from ref 15. Copyright 1982 Amencan Chemical Society.
mixture of cCcharacter (30%) from metal-based p orbitals, in addition to axial and equatorial CO x* character. To analyze the effect of these electronic transitions between highly delocalized orbitals, we employed
Trogler
242 Acc. Chem. Res., Vol. 23, No. 8, 1990 a
b
-
-
Figure 5. Difference electron density plots for the (a) loa{ 6a,’ and (b) 15e’ 6a,’ electronic transitions of R U ~ ( C Oin) the ~ ~ Ru3(CO)G plane. The large spike-like features represent noise in the subtraction of core density at Ru, which serves to mark the Ru atom locations.
a difference density analysis. Here one subtracts the total electron density distribution in the ground state from that in the excited state. The raised contours show where electron density increases after electronic excitation, and the depressed contours show where the electron density has decreased. These three-dimensional plots, shown in Figure 5, reveal a distinct difference between the two excited states. Noise wells from differences in core densities mark the location of the three Ru atoms and six CO ligands in the plane of the triangular cluster core. The important bonding regions lie at the center of the Ru3triangle and between the Ru and CO ligands. In the first case (loal’ 6ai) little disruption of electron density in the triangular core is observed, while the 15e’ 6ai transition shows a large depression in bonding electron density at the center of the Ru3 triangle. This conclusion is partially supported by the resonance Raman effect observed when the lowest dipoleallowed (15e’ 16ai) electronic transitions in Ru3(C0)12 and O~~(CO)~~[PPh~(neomenthyl)]~ were excited. Both the al’ (Ru, 185 cm-l; Os, 190 cm-l) and the Jahn-Teller-active e’ (Ru, 130 cm-l; Os, 140 cm-l) metal-metal stretching frequencies exhibit a 2-&fold intensity enhancement. Since only degenerate excited states can undergo a Jahn-Teller distortion, the enhancement of the Jahn-Teller-active mode supports an assignment of the lowest allowed electronic transition as a degenerate metal-localized Q Q* excitation. Resonance-enhanced vibrations correspond to the bonds most disturbed in the resonant excited state. Previous MCD studies also were consistent with the presence of a low-lying dipole-allowed degenerate excited state of predicted metal-metal antibonding character.20 The calculated changes in Ru-CO bonding follow an opposite pattern. In loal’ 6 a i a double valley develops (Figure 5) between the Ru and CO groups, which represents a loss in vbonding density; however, the Ru-CO bonding is relatively undisturbed (i.e., flat) for the 15e’ 16ai difference electron density plot. The presence of these two closely spaced excited states, one metal-metal antibonding and the other metal-C0 antibonding, may explain the competing metal-metal bond and metal-CO bond cleavage that occurs on photochemical excitation. This suggestion has been supported by several photochemical studies.lg Tetranuclear Metal Cluster Compounds. The complexity of tetranuclear cluster complexes leads to a further lowering of resolution in the spectroscopic
-
-
-
-
-
-
(20) Tyler, D. R.; Levenson, R. A.; Gray, H. B. J. Am. Chem. SOC.1978, 100, 7888.
1512
CO4K 0jl2
Rh,(CO),,
---
j
dband
lr4(COl,2
Figure 6. Orbital energy diagrams for M4(C0)12clusters from
SCF-Xa-DV calculations. Reprinted with permission from ‘;ef 24. Copyright 1987 American Chemical Society.
information obtained, and one relies more heavily on computational methods as a guide. Thermal instability and a lack of volatility preclude gas-phase PES measurements for most systems. A problem with the molecular orbital approach (Figure 6) is that one becomes inundated by information that provides little chemical insight. For these reasons we initially examined bare M4 clusters of iron, cobalt, and nickel.21 Molecular beam technology has permitted such compounds to be prepared and characterized spectroscopically in the gas phase.22 These naked metal clusters were predicted to be highly magnetic by the spin-polarized SCF-XaDV calculations with 12, six, and two unpaired electrons, respectively. The calculated exchange splittings of the spin-up and spin-down d orbitals also decreased (2.68,1.60, and 0.56 eV, respectively) in this series. The Fe and Ni complexes, which contain singly occupied and tl levels, should undergo Jahn-Teller distortion to C& or lower symmetry structures. Extensive mixing of metal 4s and 4p character into the metal-metal bonding orbitals was observed, which agreed with previous Hiickel calculation^.^^ Such hybridization also occurs in the coordinatively saturated C O ~ ( C Osystem. )~~ The C O ~CO)12 ( complex possesses a diamagnetic ground state, as do the Rh and Ir analogues (Figure 6).% Comparisons within this series are complicated by the (21) Holland, G. F.; Ellis, D. E.; Trogler, W. C. J. Chem. Phys. 1985, 83, 3507. (22) Andres, R. P.; Averback, R. S.; Brown, W. L.; Brus, L. E.; God-
dard, W. A., 111; Kaldor, A.; Louie, s. G.; Moscovits, M.; Peercy, P. s.; Riley, S. J.; Spaepen, F.; Wang, Y. J. Mater. Res. 1989, 4, 704. (23) Hoffmann, R.; Schilling, B. E. R.; Bau, R.; Kaesz, H. D.; Mingos, D. M. P. J. Am. Chem. SOC.1978,100,6088. Schilling, B. E. R.; Hoffmann, R. J. Am. Chem. SOC.1979,101,3456. (24) Holland, G. F.; Ellis, D. E.; Tyler, D. R.; Gray, H. B.; Trogler, W. C. J. Am. Chem. SOC.1987,109,4276.
Metal Cluster Complexes
Acc. Chem. Res., Vol. 23, No. 8,1990 243
Table I Physical Data for Tetranuclear Metal Cluster Complexes redn:" oxidn:a calcd M4 Map-Mb, Mb-Mb, complex Eo' or Epo V Eo' or EM,V core chargeb A A -0.12* +1.7* +3.11 2.492c 2.492 co4(co)i2 Co4(CO)&ripod) -0.78 +0.94 +2.41 2.540d 2.457 (?-MeCsHs)Co4(CO)e -0.59 +1.2* +2.58 2.485e 2.457 (~MeC6HdCo4(CO)s(tripod) -1.21 +0.39 +I30 2.472f 2.447 (s-C6Me6)CO4(CO)6(tripOd) -1.18 +0.45* -0.80* +5.39 2.7328 RhdC0)iz Rh&2O)&ripod) -1.14 Ir4(CO)&ripod) -1.82* 2.6&ih +5.40 2.693' 2.693 IrdC0)12
highest uco, cm-l
375 (19Ooo) 380 (11 800) 385 (11800) 380 (11 200) 385 (14300) 303 (16 600) 380 (13000) 342 (4400) 321 (7500)
2103* 205d 2078' 198@ 2101' 206om 2064" 2071O
aPotentials represent E I I Pvs an Ag wire pseudoreference at 22 "C from cyclic voltammetry; peak potentials for chemically irreversible couples are indicated by an asterisk. The potential of the ferrocene/ferrocenium couple is +0.47. Potentials are for 0.2 M tetrabutylammonium tetrafluoroborate/CHzClzsolutions, except for C O ~ ( C Oand ) ~ ~Rh4(C0),, measured in THF. Data from refs 24 and 26. *Values represent the volume-integrated charge in electrons per molecule for the four metal atoms from refs 24 and 26. Carre, F. H.; Cotton, F. A.; Frenz, B. A. Znorg. Chem. 1976,I5, 380. dDarensbourg, D. J.; Zalewski, D. J.; Delord, T. J. Organometallics 1984,3, 1210. 'Reference 28b. CBird,P. H.; Fraser, A. R. J. Organomet. Chem. 1974, 73, 103. gWei, C. H. Znorg. Chem. 1969,8, 2384. hC1ucas, J. A.; Harding, M. M.; Nicholls, B. S.;Smith, A. K. J. Chem. SOC.,Chem. Commun. 1984,319. 'Churchill, M. R.; Hutchinson, J. P. Znorg. Chem. 1978,17, 3528. 'References 24 and 26. kNoack, K. Helo. Chim. Acta 1962,45,1847. 'Reference 26. mMartinengo,S.;Giordano, G.; Chini, P. Inorg. Synth. 1980,20, 209. "Reference 54. OChaston, S. H. H.; Stone, F. G. A. Chem. Commun. 1967, 964.
change in structure from CSU(three edge-bridging CO's on a triangular face of the M4 tetrahedron) to Td(all terminal carbonyls) that occurs on progressing from Rh to Ir. With this caveat, one trend evident from the MO diagrams of Figure 6 is the increase in the energy splitting of the occupied band of d orbitals on descending the periodic table. This can be attributed in part to the increased strength of metal-metal bonding with increasing atomic number. The greatest change occurs on going from the second to the third row of the transition elements. The lowest unoccupied orbital in the M4(CO)12 clusters, 27e or 9tl, has the character of a a,* orbital with respect to bonding in a triangular face of the cluster. Thus, an intense transition in all three clusters that blue shifts on cooling to 77 K (as do u u* transitions in dinuclear metal-metal bonded carbonyls) is assigned%by analogy to the ''u u* n transition of the cluster. Spectroscopic energies for this transition are provided in Table I, along with those for derivatives where the basal face of the cluster has been capped with the tripod [HC(PPh,),] ligand. The energy of this transition is remarkably constant between 300 and 380 nm and seems to be characteristic for the M4 grouping. On the basis of the increased d-orbital splitting in the Ir cluster (Figure 6), one might have expected the transition energy to be shifted to significantly higher energies. This may be offset by the absence of bridging CO ligands in the Ir clusters, because bridging CO groups are known to shift the energy of the u u* transition to higher energy in dinuclear cobalt carbonyl complexes.26 In a study of ligand effects with the C% symmetry C O ~ clusters, substitutions were made at the basal triangular face of the cluster with the tripod ligand, at the apical Co of the cluster with an q6-arene ligand, and at both positions simultaneously.2e Because of the complexity of an orbital-by-orbital analysis of bonding in these highly delocalized clusters, we decided to approach the problem with a density of states (DOS) analysis similar to the description of bonding in solids. In this scheme each orbital in the MO diagram is assigned a 0.4-eVwide Lorentzian line shape, and the resultant s u m over
-
-
-
(25) Abrahameon, H. B.; Frazier, C. C.; Ginley, D. S.; Gray, H. B.; Lilienthal,J.; Tyler, D. R.; Wrighton, M. S. Inorg. Chem. 1977,16,1564. (26) Holland, G. F.; Ellis, D. E.; Trogler, W. C. J. Am. Chem. SOC. 1986, 108, 1884.
a
b
Figure 7. Apical- and basal-site contributions from cobalt 3d orbitals, scaled relative to the total DOS and plotted on a per-atom basis for (a) CO~(CO)~, and (b) (q-C6H6)Co4(CO),;(-1 apical; (4 basal. Reprinted with permission from ref 26. Copyright 1986 American Chemical Society.
all orbitals yields the total DOS. The real advantage in this method is that partial DOS plots can be constructed for a specific atomic or orbital contribution. For example, Figure 7a contains the apical- and basal-site Co 3d orbital contribution plotted on a per metal atom basis for C O ~ ( C Oand ) ~ ~for (~-C&6)cOq(C0)p The displacement of the peak for the prominent occupied Co 3d band below the Fermi energy (horizontal dashed line for the HOMO-LUMO gap) to higher binding energy for the basal cobalts suggests a greater electron-withdrawing ability of the bridging carbonyls in the basal plane. Substitution of the apical CO's by a benzene ligand destabilizes the apical cobalt d orbitals (Figure 7b), which agrees with the expected poor a-acceptor character of benzene as compared to CO. The greater acceptor ability of the bridging CO groups, regardless of basal or apical substitution, is apparent in other DOS plots for the CO groups where the binding-energy peak for electrons localized on the bridging CO are shifted to lower binding energy than those for the terminal CO groups. We believe the DOS approach has considerable merit for the analysis of orbital energy diagrams for complex molecular systems. Electrochemistry of Tetranuclear Cluster Compounds. Further evidence for the delocalized character of bonding in these clusters was obtained from a con-
244 Ace. Chem. Res., Vol. 23,No. 8, 1990
Trogler
primarily by a dissociative route,32p33when the entering sideration of their electrochemistry (Table I).26927In ligand, L, is a phosphorus donor. The mechanism of the C Oseries ~ the effect of replacing either the apical CO substitution in C O ~ ( C Oand ) ~ Rh4(C0)12 ~ was less or basal CO ligands with arene or tripod ligands, rewell understood. Exchange of CO in C O ~ ( C Owith )~~ spectively, results in a shift of either the first anodic 13C0 and 14C0 proceeds by a dissociative path, while or the first cathodic one-electron processes to a more fragmentation to C O ~ ( C O(at ) ~ high temperature and negative potential. Although the redox waves observed high CO pressure) follows a rate law with first- and for C O ~ ( C Owere ) ~ ~irreversible, the tripod- and arenesecond-order terms in CO.% In contrast, the reaction substituted species generally showed chemically rebetween C O ~ ( C Oand ) ~ ~L occurred too rapidly to be versible waves for the one-electron-reducedor -oxidized measured, which suggests an associative mechanism.35 radicals (Table I). Several effects may be operative. Only a lower limit could be set on the rate of CO subFirst, the tripod ligand is known to hinder fragmentastitution in Rh4(C0)12.36 Part of the motivation for tion processes in the neutral cluster?* and it may serve examining the series was the report that CO dissociation a similar role in the radical cation or anion. On elecrates of M3(CO)12 clusters follow the order3' Fe > Ru tronic and steric grounds, phosphorus donor ligand > Os. This contrasts with trends observed in periodic substitution for CO in mononuclear 17-electron comseries of mononuclear complexes where an enhanced pounds results in enhanced stability of radical cationsaB reactivity for second-row transition metals usually obArene substitution may also enhance Co(apical)-Cotains. (basal) bonding, as suggested by the bond lengths provided in Table I and by earlier extended Huckel When the reaction between Co4(CO)12and PPh, was calc~lations.~~ The potential shifts are additive, Le., examined with a combination of stopped-flow and the negative potential shift in the species with both low-temperature methods,% we found that the CO basal tripod and apical arene ligands equals the sum substitution process showed an induction period that of the corresponding shifts in the two monosubstituted varied from 1 to 15 s at 28 "C to as long as 2 h at -43 species. "C. Although this reaction may proceed by electronFrom X a calculations of (q46H6)C~4(CO)g, CO~(C- transfer catalysis, as for example in H3CCCo3(C0)9,39 O)S[(PH~)~CHI, and (a-CeHs)Co4(Co),[(pH2)3CHl, the intrinsic reactivity of Co4(CO)12is low. In contrast there was no apparent correlation between the redox the rate of CO substitution in Rh4(CO)12is well-behaved potentials and either the HOMO or LUMO orbital enand both mono- and disubstituted products form acergies. This contrasts with the correlations observed cording to a second-orderrate law k[Rh&0),,][PPh3]. in mononuclear systems.31 An excellent linear correWhile it was apparent that the reactivity of Rh4(C0)12 lation (correlation coefficient 0.997) was obtained with exceeded that of C O ~ ( C Oand ) ~ ~Ir4(CO)12,a quantitative the volume-integrated cluster-core charge for the C O ~ comparison was precluded because of the autocatalytic unit (Table I). Additivity of ligand effects, regardless nature of the C04(CO)12reactions. Therefore, the staof the site of substitution, and the correlation between bilized tripod-substituted clusters were examined. net cluster-core charges and redox potentials support The substitution of CO in M4(CO)g[HC(PPh2)3] the notion that metal clusters behave as delocalized clusters (M = Co, Rh, and Ir) proceeds with a variety sources and sinks of electrons. The lack of correlation of phosphorus donor ligands, L, to yield the apicalbetween cluster HOMO energies and redox potentials substituted M4(CO)6[HC(PPh2)3]L.Kinetic studiesq0 may result because the energy of an individual orbital, show that two-term rate laws (k, + k2[L])[M4(C0)9such as one localized on an apical cobalt, does not reflect [HC(PPhJ3]] are observed for M = Co and Rh, whereas gross molecular properties. In mononuclear complexes, for Ir the second-order term dominates for all L except which usually contain metal-localized HOMOS and 13C0. Activation parameters for the kl terms (AH*, = LUMOs, frontier orbital energies adequately represent 22-26 kcal/mol and ASS1 = 0-23 cal mol-' K-I) agree metal-centered oxidations and reduction^.^^ with a CO dissociative mechanism for the k1 process. Kinetics of CO Substitution in Metal Cluster Activation parameters for the k2 terms (AW2= 5-14 Complexes. Mechanistic data for the M4(CO)12clusters kcal/mol and ASS2= -20 to -45 cal mol-' K-') and the (M = Co, Rh, Ir) was available only for Ir4(CO)12,where dependence of rate on incoming nucleophile suggest an CO substitution occurs by a two-term rate law with a associative mechanism for CO substitution, where L dominant second-order term for phosphorus nucleoattacks an apical metal atom of the cluster. Rates of p h i l e ~ .Substitution ~~ in Ir4(CO)ll~zLz (z = 1-3) occurs CO substitution in M4(CO)g[HC(PPh2)3]are generally (27) Rimmelin, J.; Lemoine, P.; Groas, M.; de Montauzon, D.Nouv. J. Chim. 1983, 7,453. Rimmelin, J.; Lemoine, P.; Gross, M.; Bahaoun, A. A.; Osborn, J. A. Ibid. 1985,9, 181. (28) (a) Arduini, A. A.; Baheoun, A. A.; Oaborn, J. A.; Voelker, C. Angew. Chem., Int. Ed. Engl. 1980,19,1024. (b) b u n , A. A.; Oabom, J. A.; Voelker, C.; Bonnet, J. J.; Lavigne, G. Organometallics 1982, 1, 1114. (c) Bahsoun, A. A.; Osborn, J. A.; Kintzinger, J.-P.; Bird, P. H.; Siriwardane, U. Nouv. J. Chim. 1984,8, 125. (29) Trogler, W. C. Int. J. Chem. Kinet. 1987,19,1025. Trogler, W. C. In Organometallic Radical Processes; Trogler, W. C., Ed.; Elsevier: Amsterdam, 1990, pp 306-337. (30) Elian, M.; Chen, M. M. L.; Mingoa, D. M. P.; Hoffmann, R. Inorg. Chem. 1976,15, 1148. (31) Sarapu, A. C.; Fenake, R. F. Inorg. Chem. 1976,14,247. Buraten, B. E. J. Am. Chem. SOC.1982,104,1299. Bursten, B. E.; Darensbourg, D. J.; Kellogg, G. E.; Lichtenberger, D. L. Inorg. Chem. 1984,23,4361. (32) Karel, K. J.; Norton, J. R. J. Am. Chem. SOC.1974, 96,6812. Dahlinger, K.; Falcone, F.; Po€, A. J. Inorg. Chem. 1986,25,2654. Sonnenberger, D. C.; Atwood, J. D. Inorg. Chem. 1981,20,3243. Stuntz, G. F.; Shapley, J. R. J. Organomet. Chem. 1981, 213, 389.
(33) Sonnenberger, D. C.; Atwood, J. D. J. Am. Chem. SOC.1982,104, 2113. Sonnenberger, D. C.; Atwood, J. D. Organometallics 1982,1,694. Darenabourg, D. J.; Baldwin-Zuchke, B. J. J.Am. Chem. SOC.1982,104, 3906. (34) Cetini, G.; Gambino, I.; Sappa, E.; Vaglio, G. A. Ric. Sci. 1967,37, 430. Bor, G.; Dietler, U. K.; Pino, P.; Po€, A. J. J. Organomet. Chem. 1978,154,301. Keeley, D. F.; Johnaton, R. E. J . Inorg. Nucl. Chem. 1989, 11,33. Darensbourg, D. J.; Peterson, B. S.; Schmidt, R. E., Jr. Organometallics 1982, 1, 306. (35) Darenabourg, D. J.; Incorvia, M. J. Inorg. Chem. 1980,19,2585. (36) Norton,J. R.; Collman, J. P. Inorg. Chem. 1973,12, 476. (37) Po€, A. J.; Sekhar, V. C. Inorg. Chem. 1985,24, 4376. (38) Kennedy, J. R; Basolo, F.; Tmgler, W. C. Inorg. Chim. Acta 1988, 146, 75. (39) Hinkelmann, K.; Heinze, J.; Schacht, H.-T.; Field, J. S.; Vahrenkamp, H. J. Am. Chem. SOC.1989, 111, 5078 and references therein. (40) Kennedy, J. R.; Selz, P.; Rheingold, A. L.; Trogler, W. C.; Basolo, F.J . Am. Chem. SOC.1989,111, 3615.
Metal Cluster Complexes
Acc. Chem. Res., Vol. 23, No. 8, 1990 245
less than in corresponding M4(C0)12clusters, since the HC(PPh2)3 ligand donates electron density to the cluster (as compared to CO). This results in enhanced r-back-bonding to CO groups in the cluster, which raises the barrier to CO dissociation. Increased electron density at M, as well as the increased steric bulk of the tripod ligand, decreases the susceptibility of the cluster toward nucleophilic attack. The M4(CO)g[HC(PPh2)3] clusters (M = Co, Rh, Ir) show increased reactivity toward CO dissociation in the order Ir
